aactuator
READING / PRIMARY SOURCES

Evidence before adjectives.

Fourteen focused references explain the methods, parameter conventions and effects that limit realism.

Read the source that answers your question.

These are primary manufacturer or official project documents, checked against the implementation. They support methods and terminology, not the accuracy of our example motor. No vendor figures or source code are copied into these pages.

For a first pass, read FOC [1], modulation [3], motor data [7] and thermal limits [8]. The remaining references answer specific implementation questions.

[1] / CONTROL ARCHITECTURE

SPRABQ2 — Sensored FOC of Three-Phase PMSM

Texas Instruments. Read §§4–5 for transforms and the current-feedback chain. Its C2000 demonstration is not our STM32 implementation.

[2] / MACHINE EQUATIONS

PMSM — sinusoidal flux distribution

MathWorks. Check the dq equations, amplitude-invariant scaling and electromagnetic torque. Its optional loss/saturation models are not automatically present here.

[3] / MODULATION

PWM Reference Generator

MathWorks. Compare SVM, sinusoidal PWM and duty generation. Our implementation uses min/max common-mode injection, not every method supported by this block.

[4] / CONTROLLER LIMITS

Anti-Windup Control Using PID Controller Block

MathWorks. Understand back-calculation and downstream saturation. The lab tracks the final voltage-vector limit; this reference is not an assurance of stability.

[5] / POWER ELECTRONICS

SLVA504A — Power Dissipation for H-Bridge / Half-Bridge Drivers

Texas Instruments. Separate conduction and switching losses. Only the former plus a simple diode loss are modeled here; do not transfer H-bridge device counts to a three-phase bridge.

[6] / BOARD LIMITATIONS

SLVA959B — Best Practices for Board Layout of Motor Drivers

Texas Instruments. Read about current-return paths, gate loops and thermal layout. The displayed board is explanatory geometry, not evidence of these requirements being met.

[7] / PARAMETER IDENTIFICATION

Motor data and simulation

maxon. Check line/phase parameter conventions, temperature, tolerances and commutation assumptions before importing a datasheet. Our profile is synthetic, not a maxon product.

[8] / THERMAL BOUNDARIES

Continuous operation range of BLDC (EC) motors

maxon. Explains why speed-dependent iron losses reduce thermal headroom. Its motor-specific temperature limit is not a universal recommendation.

[9] / CURRENT CONVENTIONS

BLDC motor with sinusoidal commutation

maxon. Highlights current conventions and standstill winding-hotspot concerns for specific motor constructions. A single winding-temperature node cannot resolve those hotspots.

[10] / SHUTDOWN

AN4277 — PWM shutdown on STM32 MCUs

STMicroelectronics. Timer BREAK and output restart depend on configuration. Our latched no-auto-rearm policy is not a universal timer behavior or a certified safety function.

[11] / FAST PROTECTION

STM32G4 analog comparators training

STMicroelectronics. Comparator routing and blanking motivate the independent trip abstraction. No board propagation time or register-level peripheral emulation is inferred.

[12] / MECHANICAL EFFECTS

Rotational Friction

MathWorks. Distinguishes Coulomb, viscous and Stribeck friction. This lab deliberately uses only a smoothed Coulomb term plus viscous drag.

[13] / 3D MECHANICS

MuJoCo 3.3.5 computation

Google DeepMind. Mechanical state, actuation, contacts and numerical solvers. Our native-to-MuJoCo 50 μs exchange remains a separate numerical coupling contract.

[14] / BROWSER ARCHITECTURE

Interacting with compiled C/C++

Emscripten project. Explains the exported C ABI used by the Worker. WebAssembly reuses the controller; it does not itself supply a physical model or hardware validation.

Our code and our evidence are different sources.

Controller · plant, inverter and sensing · reproducible signal experiments · data provenance. Tests verify defined software contracts; validation against real hardware is a separate, unperformed step.